What remains open

Decisions to be taken

• Which substances for the second and third tracks, and in what order.
• Which functional form for the curve: the one used in the simulator is a convenience, not a model of the phenomenon.
• How to operationalise the price of relief in a way tolerable to someone in early withdrawal.
• How many measurement points, at what intervals, and for how long after last use.
• Whether to include a biological arm on cell cultures, and with which laboratory.
• How large the exposed-without-addiction group must be to yield a useful reference range.
• Who approves new collaborators joining the restricted area.

Objections we expect

"The idea that dynamical signals precede a transition has been contested." True, and the criticism is recent and serious. This is why the design does not rest on statistical instability indicators alone, but on recovery parameters defined deterministically, which are less contested.

"It is not new: dynamics are already studied." Partly true, and claims of outright novelty should be dropped. Work exists describing rhythms and amplitudes in large cohorts. What does not appear to have been done is applying response and recovery parameters to a perturbation with a certain time zero, in this population, with repeatability established first. Novelty has to be demonstrated on a precise question, not asserted.

"Measuring instead of asking is an ideological position." No, and it should not be presented that way. Asking a person how they are is not judging them. The two measures observe different things, and their disagreement is itself interesting.

"The sample is already addicted: you can say nothing about vulnerability." Correct for the first phase, and we declare it at the outset. The answer is not to dodge the objection but to accept it: the vulnerability question belongs to phase two, with its activation criterion already written.

"A 95% interval means the value lies there with 95% probability." It does not, and this was our own error, now corrected. A confidence interval is not a probability statement about the true value: it describes how the procedure behaves over many repetitions.

The next step

Before any measurement on people: verify the estimation engine on existing open data, and build the catalogue of available evidence. These are the two things that can be done now, without authorisations and without expense.